mdrun refuses to start with .cpt if nsteps is -1 in .tpr - Redmine #2757
Archive from user: Carsten Kutzner
Should be reproducible with any .tpr that has nsteps –1 and a checkpoint file.
Error was introduced in commit 4dcb2a1a “Issue fatal error if checkpoint does not suit the .tpr”. There is also a note in the added lines in checkpoint.cpp that we do not intend to support the use of mdrun -nsteps –1, but in this case the –1 comes from the .mdp file setting.
mdrun_threads_AVX2_256 -v -pin on -nt 20 -noappend -nb gpu -pme gpu -maxh 48 -s topol.tpr -cpi state_08.cpt -cpo state_09.cpt
Error message is something like
Program: mdrun_threads_AVX2_256, version 2018.4-dev-20181112-3fd2d784
Source file: src/gromacs/fileio/checkpoint.cpp (line 2622)
The input requested –1 steps, however the checkpoint file has already reached
step 188393400. The simulation will not proceed, because either your
simulation is already complete, or your combination of input files don’t
(from redmine: issue id 2757, created on 2018-11-14 by gmxdefault, closed on 2018-11-16)
- Revision 61285613 by Mark Abraham on 2018-11-15T16:37:21Z:
Fix checkpoint restart of tpr with infinite step count The recent fix of #2717 did not account for the way a user's .tpr file can require an infinite number of steps by using the special value -1. Such special values are difficult to remember when maintaining the code, so we should tend to avoid introducing them. Fixes #2757 Change-Id: I6570c4f4e7d63b2375dbb595a514c9e709f18856